[daip] Re: FRING (fwd)

[Chris Flatters]

zshen at hotaka.mtk.nao.ac.jp writes

> I did what you suggested to register the fringes on a single baseline
> which is isolated from the others. So, as you said, we have two
> "subarray"in the sense of fringefitting. What I am not so clear is
> what you said below about the inconsistency in applying the CL 
> table. Would you please tell me more about this?

It sounds as if you have 2 antennae that are not connected to the rest
of the array by any observed baseline. This is going to make data reduction
extremely difficult.

Consider a case where we have an array with antennae A, B, C, D, E and
baselines A-B, A-C, B-C, and D-E have been observed. If we choose A as
a reference antenna we can determine the delays of B and C relative to A
but not those of D and E. However, to remove the effects of the delays,
we only need to know the differences D(B) - D(A), D(C) - D(A), D(C) - D(B)
and D(E) - D(D). We can do this in 2 steps:

1 - Run FRING on A, B, and C with A as the reference antennae. Use CLCAL
    to generate a CL table using OPCODE = 'CALP'.
    
2 - Run FRING on D and E with D as the reference antenna. Use CLCAL
    to genarate a CL table using OPCODE = 'CALP' and using the CL table
    from step 1 as the input CL table (GAINVER).
    
This will give you delays D(A), D(B), D(C), D(D), and D(E) on the assumption
that D(D) = D(A). This assumption is unlikely to be correct and would lead
to incorrect delay corrections for baseline A-D, A-E, B-D, B-E, C-D, and C-E
but this will not matter since those baselines were not observed. The CL
table from step 2 can be applied to the data.

Unfortunately, we then have to deal with the fact that we have one 3 element
array and one 2 element array with no phase connection between them. Tying
these together will take some work; one possible way to proceed would be to
map the source using the A,B,C array and use the model from this to self-cal
the phases for the D,E array.

Since A, B, and C have no real connection to D and E, it will be easier to
keep track of the data if the antennae are explicitly assigned to different 
subarrays using USUBA.

> 
> I did a test on some strong compact sources which have very good
> fringes on all baselines. I deliberately single out 2 antennas to
> do fringefitting also. Then I compared the delay solutions to these
> two antennas. I found there is difference as large as 3 nanoseconds.
> If I only run FFT (No LLS) in the global fringefitting (using all the 
> antennas), the difference becomes small (about 1 nanosecond).

Why do you think that this might be important? The LLS stage is expected
to get slightly different results from the FFT search. The FFT search is
constrained to finding delays that correspond to points on the FFT grid.

> 
> I also tried baseline-based fringefitting "BLING", but got no lucky.
> It seems that "BLING" has changed a lot compared to my last try
> in 1995-6. It didn't work properly. Do you know why?

You need to explain what you mean by "didn't work properly".


	Chris Flatters
	cflatter at nrao.edu